Contrib Mineral Petrol (2009) 158:785–801 DOI 10.1007/s00410-009-0410-9
ORIGINAL PAPER
The petrogenesis of sodic island arc magmas at Savo volcano, Solomon Islands D. J. Smith Æ M. G. Petterson Æ A. D. Saunders Æ I. L. Millar Æ G. R. T. Jenkin Æ T. Toba Æ J. Naden Æ J. M. Cook
Received: 6 November 2008 / Accepted: 5 May 2009 / Published online: 28 May 2009 Ó Springer-Verlag 2009
Abstract Savo, Solomon Islands, is a historically active volcano dominated by sodic, alkaline lavas, and pyroclastic rocks with up to 7.5 wt% Na2O, and high Sr, arc-like trace element chemistry. The suite is dominated by mugearites (plagioclase–clinopyroxene–magnetite ± amphibole ± olivine) and trachytes (plagioclase–amphibole–magnetite ± biotite). The presence of hydrous minerals (amphibole, biotite) indicates relatively wet magmas. In such melts, plagioclase is relatively unstable relative to iron oxides and ferromagnesian silicates; it is the latter minerals (particularly hornblende) that dominate cumulate nodules at Savo and drive the chemical differentiation of the suite, with a limited role for plagioclase. This is potentially occurring in a crustal ‘‘hot zone’’, with major chemical differentiation occurring at depth. Batches of magma Communicated by T. L. Grove.
Electronic supplementary material The online version of this article (doi:10.1007/s00410-009-0410-9) contains supplementary material, which is available to authorized users. D. J. Smith (&) M. G. Petterson A. D. Saunders G. R. T. Jenkin Department of Geology, University of Leicester, Leicester LE1 7RH, UK e-mail:
[email protected] M. G. Petterson J. Naden J. M. Cook British Geological Survey, Keyworth, Nottingham NG12 5GG, UK I. L. Millar NERC Isotope Geosciences Laboratory, Kingsley Dunham Centre, Keyworth, Nottingham NG12 5GG, UK T. Toba Ministry of Energy Water and Mineral Resources, Honiara, Solomon Islands
ascend periodically, where they are subject to decompression, water saturation and further cooling, resulting in closed-system crystallisation of plagioclase, and ultimately the production of sodic, crystal and feldspar-rich, high-Sr rocks. The sodic and hydrous nature of the parental magmas is interpreted to be the result of partial melting of metasomatised mantle, but radiogenic isotope data (Pb, Sr, Nd) cannot uniquely identify the source of the metasomatic agent. Keywords Sodic Island arc Trachyte Mugearite Alkaline Solomon Islands Savo Hot zone
Introduction The Solomon Islands arc has experienced a complex subduction history, involving the collision of an oceanic large igneous province (the Ontong Java Plateau; Hughes 2004; Petterson et al. 1997), initiation of a second subduction zone (Petterson et al. 1999), and the subduction of a young oceanic spreading ridge (Crook and Taylor 1994). These process have contributed to a number of unusual magmatic features, including volcanism on the plate being subducted (Simbo; Ko¨nig et al. 2007) and in the forearc region (Kavachi; Johnson and Tuni 1987); picritic magmas (Schuth et al. 2004); high magnesian andesites (HMAs) and adakites (Ko¨nig et al. 2007); and alkaline magmas (DeLong et al. 1975). Savo occupies an ambiguous tectonic position, approximately equidistant between the two opposing subduction zones that define the arc (Fig. 1). The volcano is constructed on unknown basement. Fresh magmatic rocks at Savo display a number of unusual characteristics, including high sodium contents (up to 7.5 wt% Na2O), increasing Sr
123
786 Fig. 1 Map of the Solomon Islands showing major tectonic features, and locations discussed in the text. Woodlark seafloor ages from Taylor (1987). Plate motions from Petterson et al. (1999)
Contrib Mineral Petrol (2009) 158:785–801
N
100 km
Santa Isabel Simbo
NSTS
Pacific PlateOntong Java Plateau 8° S 10 cm/yr
New Georgia
Kavachi
Savo
Woodlark Basin
Malaita
Guadalcanal
SST
S
Legend Woodlark Ridge Woodlark seafloor 0-1 Ma Woodlark seafloor 1-2 Ma Woodlark seafloor 2-3 Ma Woodlark seafloor 3-4 Ma Woodlark seafloor 4-5 Ma
Indo-Australian Plate
Makira
7 cm/yr 158° E
with fractionation, and abundant ultramafic nodules. Along with Kavachi, Savo is one of only two historically active volcanoes in the arc. The role that the various tectonic elements and events have played in magma genesis at Savo remains unclear. Previous authors have related the magmatism at Savo to the subduction zone to the south (Petterson et al. 2003) but a relationship to the northern subduction zone has not been ruled out (Cooper and Taylor 1987). There is scope for interaction between both subduction zones in magmatic processes in the Solomon Islands. Previous studies have provided brief summaries of chemistry and petrology at Savo (Petterson et al. 2003; Stanton 1994). Here we provide a critical assessment of petrogenesis at Savo, important for understanding tectonic and large-scale chemical processes of the Solomon Islands in particular, and the southwest Pacific in general.
Tectonic and geological history of the Solomon Islands The Solomon Islands were formed by the ongoing convergence of the Indo-Australian and Pacific plates (Fig. 1). Southward subduction of the Pacific plate at the North Solomon Trench System (NSTS) commenced in the Eocene (Petterson et al. 1999). Sometime before 6.4 Ma, the Ontong Java Plateau (an Alaska-sized oceanic large igneous province) ‘‘choked’’ the NSTS resulting in initiation of subduction of the Indo-Australian Plate at the South Solomon Trench System (SSTS; Cooper and Taylor 1987; Petterson et al. 1999; Phinney et al. 2004). Intermittent subduction along the NSTS still occurs (Auzende et al. 1996; Cooper and Taylor 1987; Kroenke 1995). The subducted Pacific slab may still be a source of metasomatic
123
10° S
160° E
162° E
fluids and melts to the sub-arc mantle (e.g. Ko¨nig et al. 2007). Subduction of the Indo-Australian Plate at the SSTS continues at the present day. Earthquake hypocentres show that the Indo-Australian slab dips vertically to 250 km in the western region of the arc and vertically to 100 km in the eastern end of the arc (Cooper and Taylor 1987). However, in the central Solomon Islands seismicity is low magnitude, shallow and diffuse. The poor definition of the slab by seismic studies is thought to be a result of the relative warmth of the young lithosphere of the Woodlark Basin (and its subducted equivalent) compared to older, colder lithosphere to the east and west (Cooper and Taylor 1987; Mann et al. 1998). Subduction of the Woodlark Basin is considered to be a major influence on structure and magmatism in the Solomon Islands (Cooper and Taylor 1987; Johnson et al. 1987; Taylor and Exon 1987). Spreading at the Woodlark Ridge began before 5 Ma (based on magnetic lineations; Taylor 1987) and ceased at approximately 0.5 Ma (Crook and Taylor 1994). One of the Woodlark Ridge transform faults intersects the SSTS to produce a trench-trench-transform triple junction east of Simbo Island (Crook and Taylor 1994). The combined divergence and subduction of the Woodlark Ridge is believed to have led to the formation of ‘‘windows’’ in the Indo-Australian slab beneath the arc (Chadwick et al. 2009; Johnson et al. 1987; Perfit et al. 1987; Taylor 1987; Taylor and Exon 1987). The development of slab windows is believed to be responsible for forearc magmatism at Kavachi (Johnson et al. 1987), and the genesis of island arc picrites in New Georgia (Johnson et al. 1987; Perfit et al. 1987; Rohrbach et al. 2003; Schuth et al. 2004).
Contrib Mineral Petrol (2009) 158:785–801
Savo volcano Savo volcano is 35 km northwest of Honiara, the administrative capital of the Solomon Islands (9°80 S 159°490 E). The volcano has a basal diameter of 9 km and a height of approximately 1,400 m, and is constructed on *14 km thick oceanic crust (Furumoto et al. 1976; Petterson et al. 2003). The upper portion of the volcano is above sea level—Savo Island is approximately 7 km long (N–S) by 6 km wide (E–W), with a high point of 485 m. The centre of the island is marked by a 1.5-km wide, approximately 80 m deep crater. At least two heavily vegetated, small lava domes are visible in the central crater, along with a number of steeper domes in the south and southwest of the island. Streams drain from the outer crater wall in a radial pattern, dissecting the island into a series of steep-sided gorges and valleys. A number of streams originate at inland hot springs. With the exception of the steep domes in the southwest of the island, pyroclastic and reworked pyroclastic deposits are ubiquitous on the island, and include block and ash flow, debris flow, tephra fall, lahar, and surge deposits (Petterson et al. 2003). Coherent lavas (and/or intrusive bodies) are limited to the domes in the crater and southwest of the island, and discontinuous, heavily weathered exposures in valleys (Savo map in Supplementary Material).
Sampling and analytical methods Due to intense tropical weathering, in situ outcrops were rarely suitable for analytical work. Fresh samples were collected from volcaniclastic deposits, stream-cut exposures, beaches, and wherever possible, from exposed coherent lavas. Sample location map is in the electronic supplement. A number of samples were nodules within a larger body of host rock, and some of the samples collected as individual blocks (particularly the most mafic) may represent nodules separated from the host rock during transport. Weathered surfaces were removed from samples by splitter or rock saw prior to crushing. Samples were crushed to coarse chips using a hardened steel press and rendered to a powder in an agate planetary mill at the University of Leicester, with the exception of samples for Pb isotope ratio determination which were milled to a fine powder in tungsten carbide, rather than agate. Whole rock major and trace element compositions were analysed with a Philips PW1400 (samples SV1–65) or PANalytical PW4400 Axios Advanced (samples SV151–400) X-ray fluorescence spectrometer at the University of Leicester, using standard methods (Smith 2008). A range of reference materials (RMs) were used to
787
calibrate both instruments. The precision (1r) of the major element data, across a range of compositions, was estimated to be \3% for SiO2, Al2O3, Fe2O3, MgO, CaO, and Na2O, \7% for TiO2 and MnO, \10% for K2O and P2O5. For trace elements at concentrations above 10 ppm the precision was \15% for Ba; \10% for Ce, Co, Cr, Cu, Ni, Sc; and \5% for Ga, Rb, Sr, V, Zn, Zr. For trace elements at concentrations below 10 ppm, the estimates for precision were \1 ppm for Ga, Nb, Rb; \2 ppm for Co, Cu, La, Nd, Ni, Y; \4 ppm for Ba, Cr, Th, Zr. Measured values for RMs were within 1r of accepted values. There was no significant difference between results from the two spectrometers. Rare earth elements were analysed by solution ICP-MS at the British Geological Survey, Keyworth, following sodium peroxide fusion of powdered samples (Smith 2008). Accuracy was estimated from analysis of certified RMs and was typically within 10% of the accepted values. Precision (1r) was estimated to be \0.05 ppm for Ho, Lu, Pr, Tb, Tm; \0.1 ppm for Eu, Pr; \0.2 ppm for Dy, Er, Sm, Yb; \1 ppm for Ce, Gd, La, Nd. Mineral chemistry data were collected using a JEOL 8600 Superprobe at the University of Leicester (Smith 2008). Precision for electron probe analysis was calculated from counting statistics, and was generally better than ±1% for measurements[10 wt%, and better than ±5% for contents [0.5 wt%. Microprobe data tables are in the Supplementary Material. Strontium and neodymium isotope ratio determinations were carried out at the NERC Isotope Geosciences Laboratory (NIGL). Strontium and neodymium fractions were prepared using standard techniques outlined in Royse et al. (1998). Samples were analysed using a Thermo-Finnigan Triton thermal ionisation mass spectrometer. The Sr blank at the time of analysis was 111 pg. Replicate analyses of the SRM987 standard solution gave an average value of 0.710263 ± 4 (1r, n = 50). Data are reported normalised to SRM987 = 0.710250. A number of samples were leached in HCl to ascertain the effect of minor alteration; no significant difference in 87Sr/86Sr was found between leached and unleached sample fractions. The Nd blank at the time of analysis was 132 pg. Replicate analyses of the J&M standard solution gave an average value of 0.511104 ± 0.000012 (2r, n = 50). Data are reported normalised to J&M solution standard = 0.511123. Samples for the determination of Pb isotope ratios were prepared using techniques described by Kempton and McGill (2002). Pb isotope ratios were determined at NIGL using a VG Axiom, MC-ICP-MS instrument for samples spiked with Tl. All Pb isotope ratios have been corrected relative to the NBS 981 composition of Thirlwall (2002). Blanks are typically less than 200 pg, although one high blank (2 ng) was obtained during the time of analysis.
123
788
Contrib Mineral Petrol (2009) 158:785–801
Na2O + K2O (wt %)
10
Mugearites Benmoreites Trachytes
Hornblendites Clinopyroxenites Hornblende gabbros
Trachyte
Benmoreite
8
Mugearite
6
Hawaiite
4 2
Basalt
45
Basaltic Andesite Andesite
55 SiO2 (wt %)
Dacite
65
Fig. 2 Total alkalis versus silica for samples from Savo (after Le Maitre et al. 1989). Samples recalculated to 100% on a volatile-free basis. Samples are sodic (Na2O-2 [ K2O), contain less than 20% normative quartz, and are thus classified as a hawaiite–trachyte series
Results Samples from Savo are divided into two groups: the main suite consists of crystal-rich trachytes, mugearites, and occasional benmoreites, defined on the basis of total alkalis versus silica (Fig. 2), and by virtue of their sodic chemistry (Na2O-2 [ K2O). Nodules commonly occur as inclusions within main suite samples. The nodules include a wide variety of mineralogies, and occur in range of sizes, from 30 cm in diameter, to xenocrysts and micro-nodules identifiable only at thin section scale. Petrography and mineral chemistry—main suite The main suite of samples are typically crystal-rich (55– 70% crystals by volume) and porphyritic with hyalopilitic groundmass (Fig. 3a), although a small number of the mafic (hawaiite and mugearite) samples collected from small exposures of lava flow deposits are entirely crystalline, with phenocrysts of clinopyroxene and olivine (0.5– 3 mm) in a groundmass of plagioclase crystals (0.3–3 mm; Fig. 3b). Plagioclase and magnetite occur in all samples. There is a progressive change in the mafic mineral assemblage with increasing whole rock SiO2, from clinopyroxene mugearites to clinopyroxene–amphibole mugearites, amphibole benmoreites and finally amphibole–biotite trachytes (Fig. 4). Olivine is present in a small number of the more mafic samples. Anhydrite was observed in one trachyte sample (SV40). Needle-like apatite crystals can be seen in trachytic samples. Apatite crystals are more readily distinguished as inclusions within feldspar phenocrysts due to the microlite-rich nature of the groundmass.
123
Plagioclase is ubiquitous throughout the suite, and constitutes 25–35% of the sample volume of mugearites, and 40–45% of benmoreites and trachytes (Fig. 4). Crystals are typically euhedral laths in thin section at \0.2 mm to over 10 mm in rare cases, but more typically 2–3 mm. A significant quantity of fragmented crystals is also present. Normal zoning is common (Fig. 3a). Plagioclase crystals in mugearites range from An85 to An50, often within a single crystal; in benmoreites the range is from An80 to An30. Plagioclase within trachytes is commonly An40–An10, although some crystals are An75–An20. Potassium contents are low; rare outliers with [5% orthoclase (Or) may be a result of minor alteration. Occasionally plagioclase contains inclusions of amphibole and magnetite. Magnetite occurs throughout the suite as a minor phenocryst phase (1–12% by volume; Fig. 4), with crystals typically \0.3 mm. Magnetite has been observed as inclusions within all major phenocryst phases. Many magnetite crystals show well developed exsolution lamellae under reflected light. Electron microprobe analysis of magnetite frequently returned analyses with totals \90 wt%, with iron analysed as FeO (the ideal total for pure magnetite is 93%). This may be a result of an excess of the Fe2O3 component, i.e. the magnetites are approaching the maghemite endmember. Magnetite–maghemite solid solution is probably a result of subsolidus oxidation (Haggerty 1976). Clinopyroxene occurs in the hawaiites, mugearites, and one benmoreite sample (SV12; Fig. 3c), typically as well developed phenocrysts between 0.5 and 2 mm diameter, up to 30% by volume (Fig. 4). Zoning, exsolution lamellae, and reaction rims are rarely observed. Clinopyroxene crystals occasionally contain inclusions of magnetite (up to 0.3 mm, typically well developed crystals), and rarely plagioclase (\0.2 mm, poorly developed crystals). Pyroxenes from the main suite fall in a narrow compositional range that spans the augite–diopside compositional boundary; they are typically aluminian and aluminian-ferrian diopsides/augites, with a smaller proportion (\10% of analyses) of chromian diopsides and augites. There is little covariation of pyroxene chemistry with whole rock chemistry. Olivine occurs as phenocrysts (up to 2 mm diameter) in a small number of mugearitic samples, where it constitutes up to 7% by volume (Fig. 4). Crystals are typically rounded with the margins altered to iddingsite. Microprobe analyses are available only for sample SV1; analysed crystals are Fo70–80, with Ca contents of 0.15–0.20 wt%. Amphibole occurs over a wide range of whole rock SiO2 values (from *52 to *65 wt%), and after plagioclase is the most abundant phenocryst mineral. Amphibole is commonly present as well developed crystals and laths, typically crystals are between 0.5 and 1 mm but
Contrib Mineral Petrol (2009) 158:785–801
789
Fig. 3 Thin section photomicrographs samples from Savo: a typical crystal-rich trachyte (SV10) in cross polarised light; b crystalline olivine–clinopyroxene mugearite (SV1); c benmoreite with fresh clinopyroxene, and amphibole replaced by a mixture of
clinopyroxene ? magnetite ? plagioclase (SV12); d hornblendite with clinopyroxene (SV175); e hornblendite with intercumulus plagioclase and apatite in trachyte host rock (SV181); f hornblende gabbro with hornblende, clinopyroxene, and plagioclase (SV55)
occasionally over 3 mm in length. Amphibole shows strong pleochroism (a = pale green, b = pale green–yellow, c = deep green; or a = yellow, b = reddish brown, c = brown). Zoning was observed in a number of wellformed crystals. All fresh amphiboles are hornblende group (sensu lato), the majority being pargasites and magnesiohastingsites, with a smaller number of edenites and magnesiohornblendes. Significant variations in composition (e.g. Si, NaA ? KA, Mg#) occur within single samples, and even within single crystals. For example, a single crystal in SV40 was found to be normally (with minor oscillations) zoned from Mg# (Mg/Mg ? Fe) 0.79 to 0.55, as NaA ? KA varied from 0.55 to 0.9. No systematic variation in zoning was observed for the suite; normal, reverse and oscillatory zoning all occur, as do homogeneous crystals. In a number of samples, hornblende displays opaque rims, and in some (including SV12, 20, 29) the amphibole
is completely pseudomorphed by finely crystalline opaque minerals (Fig. 3c). Stanton (1994) found the replacing assemblage to be a mixture of clinopyroxene, magnetite and plagioclase with trace quartz and hematite, in his study of pseudomorphed and rimmed amphiboles from Guadalcanal and Savo. Biotite occurs in benmoreite and trachyte samples, typically as small (\0.5 mm) crystals; rare examples with diameters up to 2 mm were observed. Biotite is often deformed, and in rare examples may be seen as inclusions within large amphibole crystals. Biotite typically constitutes between 3 and 7% by volume of samples with whole rock SiO2 [ 60 wt%. Biotites are relatively magnesium rich (Mg# typically 0.6–0.7). The majority of samples analysed are part of the annite– phlogopite series (sensu Tischendorf et al. 2007), with a significant proportion sufficiently Mg-rich to be classified as phlogopite.
123
790
Contrib Mineral Petrol (2009) 158:785–801 100
Abundance (%)
An Anorthite
80
s es rite oreit ytes les h du nm rac T No Be
Mu
a ge
90
Bytownite
60
70
Labradorite 50
40
Andesine 30
20
Oligoclase 10
Albite SV32 Sv33 SV19 SV29 SV35 SV18 SV65 SV45 SV20 SV58 SV1 SV41 SV43 SV12 SV7A SV17 SV16 SV44 SV10 SV42 SV21 SV39 SV2 SV23 SV9 SV38
0
Mugearites Hyalopilitic groundmass
Ben. Feldspar Magnetite Olivine
Trachytes Clinopyroxene Amphibole Biotite
Fig. 4 Modal mineralogy of main suite thin sections as determined by point counting (minimum 750 points). Samples are ordered by increasing SiO2 contents, left to right
Petrography and mineral chemistry—nodules Nodules are subdivided on the basis of dominant mineralogy: Hornblendites (sensu lato) are composed of amphibole ([90%) with minor clinopyroxene and magnetite ± plagioclase ± apatite. Amphibole crystals are typically [2 mm, and can measure up to 3 cm. Texture varies according to mineralogy: plagioclase ± apatite-bearing samples typically have euhedral amphibole crystals, with interstitial plagioclase and apatite; plagioclase-free and plagioclase-poor samples are much more common on Savo, and are dominated by anhedral amphibole. The samples are consistent with cumulate textures, with plagioclase-free orthocumulates (Fig. 3d) and plagioclasebearing adcumulates (Fig. 3e). Amphibole compositions in nodules overlap with those from the main suite. Plagioclase in hornblendites may be more calcic than in the main suite (SV38, SV181 nodules are An80–90) but not in all cases (Fig. 5). Clinopyroxenites are dominated by clinopyroxene, with olivine, minor amphibole, and magnetite, and display orthocumulate textures. Amphibole can be seen to be replacing clinopyroxene in a number of samples, especially at the contact between the host rock and the nodule, and also occurs as an intercumulus phase, typically leading to poikilitic textures. Clinopyroxene chemistries overlap with those of the main suite, tending to be slightly more Mg-rich. From the small number of olivine analyses there is no appreciable difference in Fo% between the nodules and main suite, but the nodule olivines are less
123
Ab
Or
Fig. 5 Plagioclase compositions from unaltered samples
calcic than those of the main suite ([0.1 versus *0.2 wt% CaO, respectively). Hornblende gabbros are coarsely crystalline (1–5 mm) with plagioclase, amphibole, clinopyroxene and magnetite (Fig. 3f). Amphiboles are often blackened or partially replaced by clinopyroxene and magnetite. Hornblende gabbros are transitional between hornblendites and main suite hawaiites and mugearites in terms of mineralogy and texture. Major element chemistry Main suite samples from Savo are mildly alkaline and sodic (Na2O-2 [ K2O), and are classed as mugearites, benmoreites and trachytes, with occasional hawaiites and dacites (Fig. 2; Table 1). The majority of analysed samples are silica saturated and metaluminous, but a small number of mugearites are nepheline-normative. Major element trends for the main suite and nodules are shown in Harker variation diagrams in Fig. 6. Samples from the main suite show well defined linear relationships for all elements relative to SiO2. There is a paucity of benmoreites relative to the mugearites and trachytes; to what extent this reflects a sampling bias rather than a real lack of benmoreites in the erupted material at Savo is difficult to ascertain. Most major element oxides show relatively simple trends versus silica. The exceptions are Al2O3, which shows a weakly defined linear increase as SiO2 increases from 50 to 55 wt% SiO2¸ and plateaus at approximately 18 wt% at higher SiO2 values. P2O5 is relatively constant at 0.18–0.26 for samples with low SiO2; at SiO2 [ 62 wt% phosphorous contents decrease steadily as silica increases. Major element trends for nodules are more scattered, due to the varied mineralogy and small sample populations. Clinopyroxenite samples (SV159 and SV161) show features—e.g. low Al; high Fe—that reflect the predominance
SV33 Tuluka MUG
50.7 0.86 16.8 9.07 0.16 4.54 9.04 4.50 2.02 0.27 2.13 100.2 527 34 77 105 20 bdl 22 bdl 23 39 849 6 438 16 85 61 11 23 3.1 15 3.4 1.0 3
Sample Location Rock type
SiO2 (wt %) TiO2 (wt %) Al2O3 (wt %) Fe2O3 (wt %) MnO (wt %) MgO (wt %) CaO (wt %) Na2O (wt %) K2O (wt %) P2O5 (wt %) LOI Total Ba (ppm) Co (ppm) Cr (ppm) Cu (ppm) Ga (ppm) Nb (ppm) Ni (ppm) Pb (ppm) Rb (ppm) Sc (ppm) Sr (ppm) Th (ppm) V (ppm) Y (ppm) Zn (ppm) Zr (ppm) La (ppm) Ce (ppm) Pr (ppm) Nd (ppm) Sm (ppm) Eu (ppm) Gd (ppm) 51.0 0.72 16.5 9.31 0.14 4.78 10.0 3.74 1.63 0.22 1.09 99.2 525 27 98 88 20 bdl 24 bdl 22 32 849 7 460 20 89 68 11 23 2.9 13 3.1 1.0 3
SV19 Pogho. MUG 51.8 0.63 16.1 9.83 0.12 4.41 9.95 3.98 1.34 0.19 1.12 99.5 494 27 92 110 24 2 7 11 34 44 1215 4 305 20 55 77 11 22 2.8 12 2.8 0.9 3
SV45 Pagha. MUG 51.9 0.75 17.4 8.76 0.13 3.68 8.00 4.32 1.72 0.21 1.99 98.9 331 24 10 77 21 bdl 9 bdl 26 30 762 5 273 17 64 79 8 17 2.4 11 2.7 1.0 3
SV20 Pogho. MUG 52.4 0.68 15.3 9.31 0.15 6.88 10.2 3.71 1.35 0.19 -0.12 100.0 327 35 214 103 19 bdl 43 bdl 21 28 709 6 225 19 61 63 8 16 2.2 11 2.4 0.8 2
SV1 Rembo. MUG 58.5 0.52 17.6 5.84 0.13 1.93 6.48 5.33 1.88 0.19 0.57 99.1 425 15 29 8 22 2 bdl bdl 31 13 863 bdl 183 16 46 102 10 20 2.6 11 2.3 0.8 2
SV12 Pogho. BEN 61.5 0.42 18.5 4.11 0.10 1.48 4.16 6.23 2.27 0.17 0.58 99.5 737 9 bdl 9 25 4 bdl bdl 40 bdl 1506 5 83 10 47 136 11 21 2.6 11 2.0 0.7 2
SV17 Pagha. TRAC
Table 1 Whole rock major and trace element chemistry, and radiogenic isotope composition of main suite samples
62.9 0.44 17.9 3.67 0.09 1.81 3.68 6.84 2.36 0.24 0.30 100.2 854 15 7 21 25 2 bdl 17 42 bdl 1630 10 83 11 45 135 14 28 3.3 13 2.5 0.7 2
SV44 Lemboni TRAC 63.1 0.37 18.2 3.37 0.10 1.26 4.31 6.10 1.99 0.16 0.16 99.1 733 7 15 21 26 4 bdl 13 45 bdl 1427 6 62 12 47 136 10 19 2.4 10 2.1 0.7 2
SV10 Pogho. TRAC 64.9 0.33 18.2 2.85 0.07 1.19 3.07 7.08 2.20 0.16 0.32 100.4 893 12 7 11 24 4 bdl 15 47 bdl 1497 7 55 8 38 124 9 17 2.1 9 1.3 0.5 1
SV39 Crater TRAC 65.1 0.25 17.7 2.36 0.06 0.87 2.41 7.56 2.35 0.10 0.26 99.0 931 bdl bdl bdl 22 4 bdl 12 49 bdl 1206 6 46 7 31 127 6 11 1.3 6 1.0 0.4 1
SV2 Rembo. TRAC
65.7 0.33 18.4 2.83 0.07 1.16 2.98 7.27 2.20 0.15 0.35 101.5 889 9 bdl 10 24 4 bdl 14 44 bdl 1420 8 53 8 38 124 10 19 2.4 10 1.7 0.5 1
SV38 Crater TRAC
Contrib Mineral Petrol (2009) 158:785–801 791
123
0.13 0.8 0.19 0.5 0.08 0.6 0.10 0.70444 0.70443 0.51298
123
MUG Mugearite, BEN benmoreite, TRAC trachyte, bdl below detection limits, Pogho. Poghorovorughala, Pagha. Paghalula, Rembo. Rembokola
0.40 2.4 0.53 1.4 0.21 1.4 0.20 0.70400 0.70399 0.51297 18.451 15.518 38.273 0.45 2.8 0.56 1.6 0.22 1.9 0.24 0.70412 0.70410 0.51297 18.449 15.521 38.269 Tb (ppm) Dy (ppm) Ho (ppm) Er (ppm) Tm (ppm) Yb (ppm) Lu (ppm) 87 Sr/86Sr 87 Sr/86Sr N 143 Nd/144Nd 206 Pb/204Pb 207 Pb/204Pb 208 Pb/204Pb
0.45 2.7 0.58 1.9 0.26 1.7 0.24 0.70405 0.70404 0.51294 18.485 15.524 38.317
0.41 2.6 0.55 1.7 0.20 1.5 0.20 0.70402 0.70401 0.51296
0.46 2.8 0.62 1.7 0.26 1.7 0.27 0.70403 0.70402 0.51298 18.449 15.520 38.265
0.39 2.2 0.51 1.4 0.20 1.5 0.22 0.70413 0.70412 0.51297 18.455 15.521 38.268
0.25 1.6 0.33 0.9 0.14 1.0 0.16 0.70421 0.70420 0.51296 18.443 15.520 38.263
18.449 15.520 38.268
0.28 1.5 0.30 0.9 0.15 1.1 0.17 0.70421 0.70420 0.51296 18.445 15.520 38.266 0.25 1.5 0.27 0.9 0.12 0.9 0.13 0.70414 0.70413
0.17 1.0 0.19 0.7 0.09 0.6 0.11 0.70417 0.70416 0.51295 18.461 15.521 38.269
SV2 Rembo. TRAC SV39 Crater TRAC SV10 Pogho. TRAC SV44 Lemboni TRAC SV17 Pagha. TRAC SV12 Pogho. BEN SV1 Rembo. MUG SV20 Pogho. MUG SV45 Pagha. MUG SV19 Pogho. MUG SV33 Tuluka MUG Sample Location Rock type
Table 1 continued
0.19 1.2 0.25 0.7 0.11 0.8 0.12 0.70418 0.70417 0.51297 18.461 15.522 38.277
Contrib Mineral Petrol (2009) 158:785–801
SV38 Crater TRAC
792
of clinopyroxene, with increased olivine contents in SV159 leading to higher Mg contents (Table 2). Hornblendites SV6A, 155 and 175 cluster closely on plots of all major element oxides, with the exception of TiO2 as it is controlled by magnetite, which is a variable minor phase. Notably, the hornblendites are co-linear with main suite samples for the major elements except Al and P. Sample SV181 plots separately from the other hornblendites, with lower SiO2 (*34 wt%) and Mg, and significantly higher Ti, Fe, and P, which is consistent with increased magnetite and apatite contents. Hornblende gabbro nodules are approximately co-linear with the main suite for the elements Al, Ca, Na, K, and Fe (Fig. 6). Magnesium contents are more varied, but as displayed by the hornblendites and clinopyroxenites, Mg is susceptible to variations in modal olivine and clinopyroxene. Trace element chemistry Trace element data (Tables 1, 2) are shown as Harker variation diagrams (Fig. 7). For the most part, incompatible elements (Ba, Zr, Rb) increase and compatible elements (Co, Cr, V) decrease with increasing SiO2. A notable feature of the suite is the increase in Sr content with SiO2. Strontium and barium show weakly bimodal distributions, with a cluster of mugearite and benmoreite samples at lower concentrations, and a trachyte cluster at higher contents. For nodules, Ba, Rb, Sr, and Zr are approximately colinear with the main suite. The compatible elements show significant scatter for the nodules relative to the main suite; Cr in particular shows significant enrichment (relative to the overall trend) in the clinopyroxenites and clinopyroxene-bearing hornblendites (i.e. excluding SV181). Rare earth element chemistry All samples are enriched relative to average C1 chondrite (Tables 1, 2; Fig. 8). As SiO2 increases, the REE profiles become steeper due to progressive depletion of the MREE and HREE. The change in slope of the REE profile can be expressed as the increase in LaN/YbN and LaN/DyN from 3–4 to 6–10 as SiO2 increases from 50 to 66 wt%; over the same silica range, DyN/YbN increases from 2 to 4. Europium anomalies are absent in most samples, with only trachytes SV2, 17 and 39 showing weak positive anomalies (Eu/Eu* = 1.2, based on the deviation from the geometric mean of SmN and GdN). The rare earth profiles for two hornblendites (SV181 and SV6A) are included in Fig. 8. The two differ in terms of LREE, with SV181 considerably more enriched in La–Nd compared to SV6A. This is assumed to be a result of high
Contrib Mineral Petrol (2009) 158:785–801
CaO (wt. %)
20
15 10
1.4
7
1.0 0.6
5 3 1 2.5
K2O (wt. %)
20
Fe2O3 (wt. %)
10 5
0.2
15 10 5
1.5
0.5 1.4
P2O5 (wt. %)
25
MgO (wt. %)
15
5
Na2O (wt. %)
Al2O3 (wt. %)
20
TiO2 (wt. %)
Fig. 6 Major element Harker variation diagrams for main suite samples and nodules from Savo Island. All data recalculated to 100% on a volatile-free basis. Error bars show typical precision (one standard deviation)
793
15
5
1.0 0.6 0.2
30
40
50 60 SiO2 (wt %) Mugearites Benmoreites Trachytes
apatite content in SV181, also reflected in the high P2O5 content.
70
30
Hornblendites Clinopyroxenites Hornblende gabbros
40
50 60 SiO2 (wt %)
70
1 standard deviation
variation of 143Nd/144Nd with increasing SiO2, with data clustered tightly around 0.51297.
Radiogenic isotopes Discussion In the main suite, 87Sr/86Sr ranges between 0.7040 and 0.7042 (Table 1; Fig. 9). Trachytes tend to dominate the upper end of this range, but overlap with mugearites, so there appears to be little significant variation in 87Sr/86Sr with composition. The lead isotope ratios show no resolvable variation with increasing whole rock SiO2 (Table 1; Fig. 9), and are similar to volcanic samples from the western Solomon Islands (Ko¨nig et al. 2007). There is no observable
The role of water in petrogenesis at Savo A number of petrological and geochemical features point to high pre-eruptive water contents in the magmas at Savo. High concentrations of fluid-mobile elements (Sr, Ba, Rb) in mafic samples are consistent with melt generation from hydrated mantle (Pearce 1982), and the presence of amphibole and biotite requires high water contents
123
794
Contrib Mineral Petrol (2009) 158:785–801
Table 2 Whole rock major and trace element chemistry, and radiogenic isotope composition of nodules Sample
Standards
Location Rock type
Average
SiO2 (wt %)
39.4a
0.32
44.4
49.8
34.1
42.0
42.8
43.4
40.6
43.8
49.7
TiO2 (wt %)
3.84
0.12
0.31
0.31
1.61
1.07
0.91
0.52
1.22
1.18
0.79
Al2O3 (wt %) Fe2O3 (wt %)
8.52 18.1
0.16 0.11
3.5 13.2
3.0 8.47
13.4 22.7
11.3 13.7
10.8 13.4
12.2 12.5
18.6 14.1
18.2 13.5
17.4 8.73
MnO (wt %)
0.16
0.01
0.21
0.12
0.23
0.12
0.12
0.22
0.18
0.12
0.17
MgO (wt %)
13.9
0.12
26.3
19.7
7.92
14.4
14.8
14.2
6.26
6.59
3.10
CaO (wt %)
14.9
0.24
10.1
19.1
13.9
13.6
13.6
12.9
13.2
12.5
9.66
Na2O (wt %)
0.79
0.02
0.90
0.29
2.56
2.25
2.22
2.34
2.50
2.51
3.92
K2O (wt %)
0.20
0.02
0.17
0.03
0.68
0.54
0.58
0.46
0.44
0.56
1.77
P2O5 (wt %)
0.06
0.01
1r
LOI Total
SV159
SV161
SV181
SV155
SV175
SV6A
SV59
SV350
SV365
Soulo. CPXITE
Soulo. CPXITE
Soulo. HBLITE
Soulo. HBLITE
Soulo. HBLITE
Pogholav. HBLITE
Mbonala GAB
Kalaka GAB
Tuluka GAB
0.01
bdl
1.34
0.01
0.00
0.01
1.13
0.08
0.25
1.43
-0.09
0.47
0.22
0.28
0.30
0.74
0.35
3.80
100.5
100.7
98.9
99.1
99.6
98.9
99.0
99.5
99.6
Ba (ppm)
58
8
26
9
107
61
67
33
97
139
383
Co (ppm)
84
2
103
55
39
60
63
52
38
43
25
Cr (ppm)
527
25
1236
1871
40
249
345
731
28
bdl
11
Cu (ppm)
138
5
bdl
bdl
109
bdl
81
8
225
102
87
Ga (ppm)
18
1
6
5
21
15
12
17
23
23
22
Nb (ppm) Ni (ppm)
20 198
1 7
bdl 414
bdl 163
bdl 7
bdl 89
bdl 86
2 183
2 bdl
bdl bdl
2 7
Pb (ppm)
5
2
bdl
bdl
bdl
bdl
bdl
bdl
bdl
bdl
bdl
Rb (ppm)
10
1
4
3
4
4
4
4
5
4
23
Sc (ppm)
55
1
45
85
36
75
71
63
32
45
29
Sr (ppm)
269
2
112
52
463
261
193
179
674
667
921
Th (ppm)
2
2
bdl
bdl
bdl
bdl
bdl
5
5
bdl
bdl
V (ppm)
552
20
130
185
459
572
503
214
430
444
266
Y (ppm)
15
2
6
6
26
13
11
17
24
15
19
Zn (ppm)
207
7
76
33
97
52
41
98
74
70
67
Zr (ppm)
110
4
10
6
31
16
10
19
34
30
73
La (ppm)
5b
0
6
1
Ce (ppm)
13
0
14
4
Pr (ppm)
2.0
0.0
2.7
1.0
Nd (ppm)
10
0
16
6
Sm (ppm)
3.3
0.1
4.7
1.9
Eu (ppm) Gd (ppm)
1.1 4
0.0 0
1.6 0.75
0.6 0.39
Tb (ppm)
0.68
0.01
6
2
Dy (ppm)
4.9
0.1
4.7
2.3
Ho (ppm)
1.01
0.02
0.96
0.51
Er (ppm)
3.1
0.1
2.8
1.4
Tm (ppm)
0.43
0.01
0.37
0.20
Yb (ppm)
3.0
0.2
2.4
1.2
Lu (ppm)
0.43
0.01
0.32
87
0.704110c
0.000005
Sr/86Sr
87
Sr/86Sr N Nd/144Nd
0.704097
0.70418
143
0.511104d
0.000006
206
18.380e
0.001
Pb/204Pb
123
0.24 0.70420
Contrib Mineral Petrol (2009) 158:785–801
795
Table 2 continued Sample
Standards
Location Rock type
Average
1r
207
15.563
0.001
208
38.697
0.004
Pb/204Pb Pb/204Pb
SV159
SV161
SV181
SV155
SV175
SV6A
SV59
SV350
SV365
Soulo. CPXITE
Soulo. CPXITE
Soulo. HBLITE
Soulo. HBLITE
Soulo. HBLITE
Pogholav. HBLITE
Mbonala GAB
Kalaka GAB
Tuluka GAB
Includes data (average values and one standard deviation) of reference materials analysed along with samples from Savo CPXITE Clinopyroxenite, HBLITE hornblendite, GAB hornblende gabbro, bdl below detection limits, Soulo. Soulomata catchment, Pogholav. Pogholavka a
XRF (majors and traces) standard is MRG-1 (Canadian Certified Reference Materials Project)
b
REE standard is JA-1 (Geological Survey of Japan)
c
JB-1 (Geological Survey of Japan) JB-1
Fig. 7 Selected trace element Harker variation diagrams for main suite and nodule samples from Savo Island. SiO2 from major element analyses, recalculated to 100% on a volatile-free basis. Error bars show typical precision (one standard deviation)
800
1400
Sr (ppm)
J&M solution standard
e
Ba (ppm)
d
600 400 200
1000 600 200
500
V (ppm)
Co (ppm)
100
60
20
300 100
25
Y (ppm)
Rb (ppm)
50
30 10
15 5
40
50
60
70
SiO2 (wt %) Mugearites Benmoreites Trachytes
Hornblendites Clinopyroxenites Hornblende gabbros 1 standard deviation
140
Zr (ppm)
30
100 60 20 30
40
50
60
70
SiO2 (wt %)
([3 wt%) in the crystallising magmas (Barclay and Carmichael 2004; Gill 1981; Moore and Carmichael 1998; Sisson and Grove 1993). Pre-eruptive conditions (temperature, pressure, pH2O) are difficult to determine precisely due to the mineralogy (e.g. one pyroxene, no
ilmenite), lack of glass and widespread development of opaque-rich rims on amphiboles. However, numerous experiments have been undertaken on arc basalts and the qualitative aspects of those experiments should apply to Savo.
123
Contrib Mineral Petrol (2009) 158:785–801
10
0.5134
A
Pacific MORB New Georgia
Nd/144Nd
Mugearites Benmoreite Trachytes SV181 SV6A
100
143
Abundance / average C1 chondrite
796
0.5130
Sediments Indian MORB
0.5126
0.5122 0.702
0.703
1
87
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
15.7
Fig. 8 Chondrite-normalised REE plot for a subset of samples from Savo. Normalising values from Boynton (1984). Hornblendite sample SV181 has a significant quantity of apatite (P2O5 [ 1 wt%) and is LREE-enriched relative to hornblendite SV6A
123
0.705
0.706
0.707
Sr/86Sr
B
15.6
Pacific MORB
207
Pb/204Pb
Sediments
Indian MORB
15.5
15.4 Sediments
38.8
Pb/204Pb
38.6 38.4 Pacific MORB
38.2 Indian MORB
208
High water contents in basaltic melts have the effect of destabilising silicate minerals, plagioclase in particular (Gaetani et al. 1993; Gill 1981; Sisson and Grove 1993). Plagioclase is suppressed at high water contents, and occurs as a liquidus phase at lower temperatures and pressures in hydrous basalts than in anhydrous equivalents (Gaetani et al. 1993). Oxides are affected much less than silicates by high water contents. Thus, arc basalts typically fractionate assemblages (in order of appearance, rather than abundance) of magnetite [ olivine [ clinopyroxene [ plagioclase (Sisson and Grove 1993). High fO2 (the presence of anhydrite in SV40 indicates fO2 C NNO ? 1; Carroll and Rutherford 1987) would also stabilise oxides relative to silicates. Amphibole is stabilised as a liquidus phase at high water contents ([3 wt%), particularly in sodic melts (Sisson and Grove 1993). The presence of clinopyroxenite and in particular, hornblendite cumulates, at Savo is consistent with the fractionation of a hydrous magma. With reference to the mineralogy of the main suite, clinopyroxenites would be the dominant cumulate early in the fractionation, as clinopyroxene and olivine are the main mafic phases in the least evolved mugearites (Fig. 4). Hornblende does not appear until greater degrees of differentiation, and so the hornblendites are expected to represent a later fractionating assemblage. However, the role of plagioclase in the differentiation of the suite is unclear—the majority of cumulates are plagioclase-poor, and Sr behaves as an incompatible element with increasing differentiation (Fig. 7) indicating a limited role for plagioclase in fractionating assemblages. Strontium data from main suite plagioclase crystals indicate further complexity. Figure 10a shows that the plagioclase compositions can be divided into two groups within the main suite—mugearites and benmoreites
0.704
38.0 37.8 37.6 18.0
18.4 206
18.8 204
Pb/ Pb
Fig. 9 a Sr–Nd diagram for samples from Savo. Fields for Pacific and Indo-Australian MORB from Hofmann (1997), New Georgia field from Schuth et al. (2004), and sediments field from Ko¨nig et al. (2007). Error within point size. b Whole rock common lead isotopes for Savo. Fields from Peate et al. (1997). Error within point size
(characterised by a relatively flat trend of SrO with decreasing XAn) and trachytes (sharply decreasing from high initial SrO with decreasing XAn). A small number of
Contrib Mineral Petrol (2009) 158:785–801
Mugearites Benmoreites Trachytes Typical Error (2σ)
0.7 0.6
Mugearites Benmoreites Trachytes Typical Error (2σ)
A 0.2
BaO (wt %)
0.8
SrO (wt %)
797
0.5 0.4 0.3
0.1
0.2 0.1
0
0
1.0
B
0.6
0.4
0.2
0
X An
1000
Fig. 11 Behaviour of BaO (wt%) with decreasing XAn in plagioclase for a subset of electron microprobe analyses (includes core-to-rim analysis of zoned crystals. Points are shaded according to host rock composition. Analytical precision is too low to resolve significant variations in calculated equilibrium melt compositions (not shown)
800
Sr (ppm)
0.8
600 400 200 0 1.0
0.8
0.6
0.4
0.2
0
XAn Fig. 10 a Behaviour of SrO (wt%) with decreasing XAn (mole fraction anorthite) in plagioclase for a subset of electron microprobe analyses (includes core-to-rim analysis of zones crystals). Points are shaded according to host rock composition. b Equilibrium melt Sr (ppm) contents, based on DSr values calculated for a given XAn from the equation of Blundy and Wood (1991)
benmoreite-hosted crystals lie closer to the trachyte group (i.e. elevated SrO for a given XAn). Core-to-rim traverses did not locate any elevated SrO zones with crystals; rather, the entire crystal would be elevated relative to the predominant trend. Given these crystals occur within the same sections as ‘‘normal’’ plagioclase, they may represent xenocrysts or products of magma mingling. By comparison, BaO from the same analysed crystals does not show two distinct groupings—instead it shows one trend with increasing BaO with decreasing XAn (Fig. 11). Figure 10b uses the equation of Blundy and Wood (1991) for DSr varying with XAn to calculate equilibrium melt Sr. Equilibrium melt compositions slope steeply down from high initial Sr (*1,000 ppm) as XAn decreases, within both the mugearite/ benmoreite group and trachyte groups. Closed-system crystallisation of plagioclase is the most obvious explanation of the trend within a group—plagioclase crystallises, progressively exhausting the melt of Sr. Plagioclase fractionation is limited, at least within the trachytes, as whole rock Sr is greater than any calculated melt composition. The whole rock chemistry of these crystal-rich rocks does not reflect a liquid line of descent.
Figure 10b shows that trachytes evolved from melts with higher Sr than mugearites and benmoreites. However, the BaO–XAn plot (Fig. 11) does not show two trends, and when considering that 87Sr/86Sr shows little difference between mugearitic and trachytic compositions, this would seem to rule out a significant role for assimilation. The whole rock chemistry, although not representing liquid lines of descent, does show linear trends for most elements, implying a common origin for mugearites and trachytes. Hydrous arc magmas may fractionate and evolve according to the ‘‘hot zone’’ model of Annen et al. (2006), where parental magmas undergo significant amounts of differentiation in lower crustal intrusions, with periodic release of fractionated daughter magmas into shallow chambers. To apply this model to Savo, magnetite, clinopyroxene, amphibole and limited amounts of plagioclase (plus minor olivine and apatite) are fractionated extensively in deeper intrusions (and recovered as cumulate nodules), then evolved magmas ascend to shallow levels where plagioclase is ‘‘frozen in’’ by rapid crystallisation. The primary control on chemistry is the deep fractionation event, where linear trends in elements such as Co, V, Ba, and Zr are generated for the main suite as they are either entirely compatible with the mafic mineral assemblages, or entirely incompatible (i.e. bulk D values remain above or below 1 regardless of the minerals fractionating). Strontium behaves as an incompatible element when plagioclase crystallisation is limited. Upon ascent, batches of magma are subject to decreasing pressure, decreasing water solubility, and ultimately water saturation and degassing. Plagioclase will begin crystallising, or will constitute an increasingly large proportion of the crystallising assemblage as the system moves to lower water contents, and Sr
123
798
will behave compatibly and leave the melt fraction progressively depleted (as per Fig. 10b). Viscosity will increase, thus limiting plagioclase fractionation, and so crystal-rich, high-Sr mushes will be produced. The role of the deep fractionation of mafic minerals, and in particular amphibole, has a number of important controls on geochemistry at Savo. Rare earth element profiles of the main suite exhibit curves characteristic of hornblende fractionation, with progressive MREE–HREE depletion (Fig. 8; compare with hornblendite SV6A). LREE remain relatively consistent, a result of apatite saturation and fractionation in the more evolved rocks (see P2O5-rich hornblendite SV181, Fig. 8). Hornblende fractionation (plus minor plagioclase removal in hornblende gabbro-type cumulates) also limits the progressive increase in whole rock Al2O3, preventing the melts from becoming corundum normative, and increases SiO2, moving compositions away from silica-undersaturated to saturated conditions (Martin 2007). Minimal plagioclase fractionation at Savo, and the dominance of calcic compositions in cumulates when it does occur (Fig. 5) serve to maintain the high sodium content of the whole rocks. Sodium is retained along with Sr in the feldspar, whereas in the case of plagioclase fractionation, the suite would become more potassic with differentiation. Potassium does indeed increase at Savo, but whole rock values remain sufficiently sodic that Na2O-2 is greater than K2O throughout the main suite. The trachytes of Savo show a number of geochemical characteristics in common with adakites (Defant and Drummond 1990). However, the geochemical criteria for defining a rock as adakitic are genetically non-unique (Richards and Kerrich 2007). Already the importance of amphibole fractionation has been demonstrated at Savo, and the late crystallisation and retention of plagioclase further emphasises the high Sr/Y value of the magma. The most likely explanation for the ‘‘adakitic signatures’’ at Savo is that initially Sr-rich, hydrous arc basalt/hawaiite fractionates an amphibole-dominated assemblage, resulting in increasing La/Yb, and Sr/Y. The plagioclase crystal chemistry plots show a bimodal distribution within the main suite, with only a minority of crystals from benmoreite hosts displaying intermediate SrO–XAn compositions (Fig. 10). To what extent this gap is real or an artefact of sampling is unclear. Compositional gaps can develop by periodic extraction of fractionated liquids from magmas when a critical degree of crystallisation is reached (Brophy and Dreher 2000). By this mechanism, extraction of a trachytic liquid from crystallising mugearite at Savo would require release of the daughter liquid prior to significant plagioclase crystallisation (to maintain high Sr in the daughter). Alternatively, benmoreites at Savo might have a tendency to
123
Contrib Mineral Petrol (2009) 158:785–801
remain in the upper crust, rather than ascending to surface, perhaps as a result of neutral buoyancy, high viscosity or deeper water saturation (by virtue of higher water contents than mugearites through fractionation, and lower water solubility than trachytes by virtue of melt composition). Simply put, a wide compositional range of melts can be released from depth following fractionation from a parental melt, and all begin closed-system (plagioclase-dominated) crystallisation upon water saturation—the depth of water saturation will vary for the different melt compositions. Without data to constrain melt H2O contents at Savo, this hypothesis cannot presently be tested. Sodic magmas at Savo Sodic magmas (where Na2O-2 [ K2O) are unusual occurrences in arcs, although they may be generated by a number of mechanisms, including: partial melting of subducted slabs; partial melting of underplated basaltic crust (Atherton and Petford 1993); assimilation of basaltic lower crustal material (Feeley and Hacker 1995); partial melting of mantle metasomatised by aqueous fluids (Kamenov et al. 2008; McInnes and Cameron 1994) and/or melts from subducted slabs (Kepezhinskas et al. 1995); and small degrees of partial melting of the mantle in truncated melt columns (Hole and Saunders 1996). Models involving extensive interaction with crustal material are unlikely at Savo. Radiogenic isotope data preclude the involvement of compositionally (or temporally) distinct material, and crustal thickness beneath Savo (approximately 14 km; Petterson et al. 2003) is much thinner than in those models that involve extensive crustal interaction (Atherton and Petford 1993; Feeley and Hacker 1995). DeLong et al. (1975) compiled data for sodic magmas in intra-oceanic arcs and observed that these rock types occur in a number of specific tectonic settings: (1) near lateral edges of subduction zones where hinge faulting is occurring (Bering, Fiji, Grenada) and (2) where fracture zones and ridges are being subducted at a high angle (Kanaga, Aoba and Ambrym, New Georgia, Iwo-jima). DeLong et al. (1975) suggested that the subduction of these linear features provides pathways for magmas from regions beneath or within the subducted lithosphere, an idea further expanded into a model of slab window development. Thorkelson and Breitsprecher (2005) predicted that melts above such a slab window would likely be adakitic, due to thermal erosion of the slab window margins during mantle upwelling. The slab window model is viable for the Solomon Arc, and explains a number of features in the western portion of the arc, including forearc volcanism (Johnson et al. 1987), island arc picrites (Schuth et al. 2004), and
Contrib Mineral Petrol (2009) 158:785–801
volcanism on the plate being subducted (Chadwick et al. 2009; Ko¨nig et al. 2007). Slab melts and aqueous fluids from subducted slabs can enrich the mantle by metasomatism (Gregoire et al. 2001; Kepezhinskas et al. 1995; McInnes et al. 2001; Pearce and Peate 1995; Rapp et al. 1999). At low melt/rock ratios, slab melts enrich the mantle wedge with mobile elements, and particularly with Na (Kepezhinskas et al. 1995; Xiong et al. 2006), rather than ascending to the surface as pristine adakites or mantle-hybridised HMAs (Kepezhinskas et al. 1995; Rapp et al. 1999). Given the presence of HMAs in the western Solomon Islands (New Georgia Province and Simbo; Ko¨nig et al. 2007), as well as the favourable tectonic setting for partial melting of the subducted slab, this is an appealing agent for enriching mantle-derived melts with Na at Savo. Enrichment of the mantle has occurred further west in the Tabar–Lihir–Tanga–Feni island arc of Papua New Guinea, where alkaline eruptive suites commonly contain mantle xenoliths that indicate widespread metasomatism beneath the arc (Gregoire et al. 2001; Kamenov et al. 2008; McInnes and Cameron 1994). Various interpretations have been made of the origin of the metasomatism in this region, including partial melting of subducted crust as the geotherms recover at the stalled slab, resulting in adakite genesis (Kamenov et al. 2008). However, other studies have concluded that the metasomatic agent was a hydrous fluid (Gregoire et al. 2001). Under high pressure and temperature conditions (1,250°C, 15–25 kb), aqueous fluids derived from a dehydrating slab are capable of carrying significant mass fractions of mobile elements. Under such conditions, aqueous fluids and silicate melts may be entirely miscible, and would have similar solvent properties (Ayers and Eggler 1995; Bureau and Keppler 1999). It therefore makes it difficult to discriminate between slab melt and hydrous metasomatism on the basis of trace and major elements alone. In fact, the presence of arc-like Ba and Sr enrichments in basalts erupted in the Woodlark Basin suggests that the Pacific slab has a ‘‘hydrous footprint’’ that extends a considerable distance south of the arc (Perfit et al. 1987; Woodhead et al. 1998). The geographical location of Savo places it in an ambiguous position above both the Pacific and Indo-Australian slabs. Both slabs have the potential to dehydrate, or partially melt, given the unusual tectonic scenario (stalled subduction, young crust, and slab window development). Radiogenic isotopes do not clearly distinguish fluid/melt sources at Savo, particularly if subducted sediments are involved (Fig. 9). Both processes could occur in tandem, given the two subducted slabs present beneath the arc, to give rise to alkali-enriched, metasomatised mantle and ultimately to sodic magmas at Savo.
799
Conclusions The erupted material at Savo spans a broad range of compositions, from mafic to felsic, with abundant nodules of cognate ultramafic material. The trace element, isotopic and mineral chemistry of the samples analysed in this study indicate that the various compositions at Savo have a common origin, but have undergone distinct differentiation and crystallisation histories. Fractionation at Savo is consistent with the ‘‘hot zone’’ model of arc magmatism (Annen et al. 2006), where primitive, hydrous mantle melts are emplaced at deep crustal levels, undergo high temperature and pressure differentiation by the fractionation of ferromagnesian silicates and oxides to establish major chemical variations. High water contents suppress plagioclase relative to mafic minerals. Evolved magmas ascend to shallower levels, where water saturation and further crystallisation occurs. At Savo, plagioclase crystallisation occurs as a closed-system process, resulting in strongly normal-zoned feldspars, and high whole rock strontium concentrations. Distinct feldspar crystallisation trends are visible within the suite, indicating that distinct batches of magma are released after relatively plagioclase-poor fractionation at depth, to crystallise upon ascent. Limited plagioclase fractionation retains the sodic chemistry in terms of whole rock compositions. The hydrous, sodic magmas that occur at Savo are most likely derived from a metasomatised mantle wedge. The metasomatic ‘‘footprint’’ of either the Pacific or IndoAustralian slabs extends a considerable distance away from the respective trenches, and Savo’s location equidistant between the two slabs means that an unequivocal slab source cannot be identified. The unusual tectonic features of the Solomon Islands are important contributors to forearc volcanism and picrite petrogenesis elsewhere in the arc. This study of Savo volcano indicates that alkaline magmatism is another notable product of the complex geodynamic setting of the arc. Acknowledgments This research was funded by the Natural Environment Research Council (UK) and British Geological Survey University Funding Initiative as part of Ph.D. studentship NER/S/A/ 2004/12339. Analytical work was supported by NERC Isotope Geosciences Facilities Steering Committee grant IP/889/1105. Additional fieldwork funds were provided by Society of Economic Geologists Student Research Grants (Hugh E. McKinstry Fund and Newmont Mining Corporation Grant). Thanks to K.A. Green (BGS) and A. Sumner (NIGL) for assistance with analytical work; W. Satokana, G. Albert, A. Ramo, D. Billy, and S. Basi (Solomon Islands Ministry of Natural Resources), the British High Commission (Honiara) for assistance with fieldwork; D. Selles (Harvard) for helpful discussion. We thank G. Moore, editor T.L. Grove and an anonymous reviewer for comments which have greatly improved this manuscript. M.G. Petterson, I.L. Millar, J. Naden, and J.M. Cook publish with the permission of the Executive Director, British Geological Survey (NERC).
123
800
References Annen C, Blundy JD, Sparks RSJ (2006) The genesis of intermediate and silicic magmas in deep crustal hot zones. J Petrol 47(3):505– 539. doi:10.1093/petrology/egi084 Atherton MP, Petford N (1993) Generation of sodium-rich magmas from newly underplated basaltic crust. Nature 362(6416):144– 146. doi:10.1038/362144a0 Auzende J-M, Kroenke LW, Collot JY, Lafoy Y, Pelletier B, Collot J-Y (1996) Compressive tectonism along the eastern margin of Malaita Island (Solomon Islands). Mar Geophys Res 18:289– 304. doi:10.1007/BF00286082 Ayers JC, Eggler DH (1995) Partitioning of elements between silicate melt and H2O—NaCl fluids at 1.5 and 2.0 GPa pressure: implications for mantle metasomatism. Geochim Cosmochim Acta 59(20):4237–4246. doi:10.1016/0016-7037(95)00244-T Barclay J, Carmichael ISE (2004) A hornblende basalt from western Mexico: water-saturated phase relations constrain a pressure– temperature window of eruptibility. J Petrol 45(3):485–506. doi: 10.1093/petrology/egg091 Blundy JD, Wood BJ (1991) Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase feldspar, silicate melts, and hydrothermal solutions. Geochim Cosmochim Acta 55:193–209. doi:10.1016/0016-7037(91)90411-W Boynton WV (1984) Geochemistry of the rare earth elements: meteorite studies. In: Henderson P (ed) Rare earth element geochemistry. Elsevier, Amsterdam, pp 63–114 Brophy JG, Dreher ST (2000) The origin of composition gaps at South Sister volcano, central Oregon: implications for fractional crystallization processes beneath active calc-alkaline volcanoes. J Volcanol Geotherm Res 102(3–4):287–307. doi:10.1016/ S0377-0273(00)00192-X Bureau H, Keppler H (1999) Complete miscibility between silicate melts and hydrous fluids in the upper mantle: experimental evidence and geochemical implications. Earth Planet Sci Lett 165(2):187–196. doi:10.1016/S0012-821X(98)00266-0 Carroll MR, Rutherford MJ (1987) The stability of igneous anhydrite: experimental results and implications for sulfur behavior in the 1982 El Chichon trachyandesite and other evolved magmas. J Petrol 28(5):781–801 Chadwick J, Perfit M, McInnes B, Kamenov G, Plank T, Jonasson I, Chadwick C (2009) Arc lavas on both sides of a trench: slab window effects at the Solomon Islands triple junction, SW Pacific. Earth Planet Sci Lett 279(3–4):293–302. doi:10.1016/ j.epsl.2009.01.001 Cooper P, Taylor B (1987) The spatial distribution of earthquakes, focal mechanisms and subducted lithosphere in the Solomon Islands. In: Taylor B, Exon NF (eds) Marine geology, geophysics and geochemistry of the Woodlark Basin–Solomon Islands, vol 7. Circum-Pacific Council, Houston, pp 67–80 Crook KAW, Taylor B (1994) Structure and quaternary tectonic history of the Woodlark Triple Junction region, Solomon Islands. Mar Geophys Res 16(1):65–89. doi:10.1007/BF01812446 Defant M, Drummond MS (1990) Derivation of some modern arc magmas by melting of young subducted lithosphere. Nature 347(6294):662–665. doi:10.1038/347662a0 DeLong SE, Hodges FN, Arculus R (1975) Ultramafic and mafic inclusions, Kanaga Island, Alaska, and the occurrence of alkaline rocks in island arcs. J Geol 83:721–736 Feeley TC, Hacker MD (1995) Intracrustal derivation of Na-rich andesitic and dacitic magmas—an example from Volcan Ollague, Andean Central Volcanic Zone. J Geol 103(2):213–225 Furumoto AS, Webb JP, Odegard ME, Hussong DM (1976) Seismic studies on the Ontong Java Plateau, 1970. Tectonophysics 34(1–2):71–90. doi:10.1016/0040-1951(76)90177-3
123
Contrib Mineral Petrol (2009) 158:785–801 Gaetani GA, Grove TL, Bryan WB (1993) The influence of water on the petrogenesis of subduction related igneous rocks. Nature 365(6444):332–334. doi:10.1038/365332a0 Gill JB (1981) Orogenic andesites and plate tectonics. Springer, New York Gregoire M, McInnes BIA, O’Reilly SY (2001) Hydrous metasomatism of oceanic sub-arc mantle, Lihir, Papua New Guinea—part 2. Trace element characteristics of slab-derived fluids. Lithos 59(3):91–108. doi:10.1016/S0024-4937(01)00058-5 Haggerty SE (1976) Oxidation of opaque mineral oxides in basalts. In: Rumble DIII (ed) Oxide minerals, vol 3. Mineralogical Society of America, Washington DC Hofmann AW (1997) Mantle geochemistry: the message from oceanic volcanism. Nature 385(6613):219–229. doi:10.1038/385219a0 Hole MJ, Saunders AD (1996) The generation of small melt-fractions in truncated melt columns: constraints from magmas erupted above slab windows and implications for MORB genesis. Mineral Mag 60:173–189. doi:10.1180/minmag.1996.060.398.12 Hughes GW (2004) Accretion of the Ontong Java Plateau to the Solomon arc: a historical perspective. Tectonophysics 389(3– 4):127–136. doi:10.1016/j.tecto.2004.02.013 Johnson RW, Tuni D (1987) Kavachi, an active forearc volcano in the western Solomon Islands: reported eruptions between 1950 and 1982. In: Taylor B, Exon NF (eds) Marine geology, geophysics and geochemistry of the Woodlark Basin–Solomon Islands, vol 7. Circum-Pacific Council, Houston, pp 89–112 Johnson RW, Jaques AL, Langmuir CH, Perfit MR, Staudigel H, Dunkley PN, Chappell BW, Taylor SR, Baekisapa M (1987) Ridge subduction and forearc volcanism; petrology and geochemistry of rocks dredged from the western Solomon Arc and Woodlark Basin. In: Taylor B, Exon NF (eds) Marine geology, geophysics and geochemistry of the Woodlark Basin–Solomon Islands, vol 7. Circum-Pacific Council, Houston, pp 155–226 Kamenov GD, Perfit MR, Mueller PA, Jonasson IR (2008) Controls on magmatism in an island arc environment: study of lavas and sub-arc xenoliths from the Tabar–Lihir–Tanga–Feni island chain, Papua New Guinea. Contrib Mineral Petrol 155(5):635– 656. doi:10.1007/s00410-007-0262-0 Kempton PD, McGill R (2002) Procedures for the analysis of common lead at the NERC Isotope Geosciences Laboratory and an assessment of data quality. NERC Isotope Geosciences Laboratory, Nottingham, p 59 Kepezhinskas PK, Defant MJ, Drummond MS (1995) Na metasomatism in the island-arc mantle by slab melt-peridotite interaction: evidence from mantle xenoliths in the north Kamchatka arc. J Petrol 36(6):1505–1527 Ko¨nig S, Schuth S, Mu¨nker C, Qopoto C (2007) The role of slab melting in the petrogenesis of high-Mg andesites: evidence from Simbo Volcano, Solomon Islands. Contrib Mineral Petrol 153(1):85–103. doi:10.1007/s00410-006-0136-x Kroenke LW (1995) A morphotectonic interpretation of SOPACMAPS 1:500 000 charts, Central Solomon Islands—Southern Tuvalu. In: SOPAC Technical Report. SOPAC, Fiji Le Maitre RW, Bateman P, Dudek A, Keller J, Lameyre J, Le Bas MJ, Sabine PA, Schmid R, Sorensen H, Streckeisen A, Woolley AR, Zanettin B (1989) A classification of igneous rocks and glossary of terms: recommendations of the international union of geological sciences subcommission on the systematics of igneous rocks. Blackwell Scientific, Oxford Mann P, Taylor FW, Lagoe MB, Quarles A, Burr G (1998) Accelerating late quaternary uplift of the New Georgia Island Group (Solomon Island arc) in response to subduction of the recently active Woodlark spreading center and Coleman seamount. Tectonophysics 295(3–4):259–306. doi:10.1016/S00401951(98)00129-2
Contrib Mineral Petrol (2009) 158:785–801 Martin RF (2007) Amphiboles in the igneous environment. In: Hawthorne FC et al (eds) Amphiboles: crystal chemistry, occurrence, and health issues, vol 67. Mineralogical Society of America, Chantilly, pp 323–357 McInnes BIA, Cameron EM (1994) Carbonated, alkaline hybridizing melts from a sub-arc environment: mantle wedge samples from the Tabar–Lihir–Tanga–Feni arc, Papua New Guinea. Earth Planet Sci Lett 122(1–2):125–141. doi:10.1016/0012-821X(94) 90055-8 McInnes BIA, Gregoire M, Binns RA, Herzig PM, Hannington MD (2001) Hydrous metasomatism of oceanic sub-arc mantle, Lihir, Papua New Guinea: petrology and geochemistry of fluidmetasomatised mantle wedge xenoliths. Earth Planet Sci Lett 188(1–2):169–183. doi:10.1016/S0012-821X(01)00306-5 Moore G, Carmichael ISE (1998) The hydrous phase equilibria (to 3 kbar) of an andesite and basaltic andesite from western Mexico: constraints on water content and conditions of phenocryst growth. Contrib Mineral Petrol 130(3–4):304–319. doi: 10.1007/s004100050367 Pearce JA (1982) Trace element characteristics of lavas from destructive plate boundaries. In: Thorpe RS (ed) Andesites: orogenic andesites and related rocks. Wiley, Chichester, pp 525– 548 Pearce JA, Peate DW (1995) Tectonic implications of the composition of volcanic arc magmas. Annu Rev Earth Planet Sci 23:251– 285. doi:10.1146/annurev.ea.23.050195.001343 Peate DW, Pearce JA, Hawkesworth CJ, Colley H, Edwards CMH, Hirose K (1997) Geochemical variations in Vanuatu arc lavas: the role of subducted material and a variable mantle wedge composition. J Petrol 38(10):1331–1358. doi:10.1093/petrology/ 38.10.1331 Perfit MR, Langmuir CH, Baekisapa M, Chappell BW, Johnson RW, Staudigel H, Taylor SR (1987) Geochemistry and petrology of volcanic rocks from the Woodlark Basin; addressing questions of ridge subduction. In: Taylor B, Exon NF (eds) Marine geology, geophysics and geochemistry of the Woodlark Basin–Solomon Islands, vol 7. Circum-Pacific Council, Houston, pp 113–154 Petterson MG, Neal CR, Mahoney JJ, Kroenke LW, Saunders AD, Babbs TL, Duncan RA, Tolia D, McGrail B (1997) Structure and deformation of north and central Malaita, Solomon Islands; tectonic implications for the Ontong Java Plateau-Solomon arc collision, and for the fate of oceanic plateaus. Tectonophysics 283(1–4):1–33. doi:10.1016/S0040-1951(97)00206-0 Petterson MG, Babbs T, Neal CR, Mahoney JJ, Saunders AD, Duncan RA, Tolia D, Magu R, Qopoto C, Mahoa H, Natogga D (1999) Geological-tectonic framework of Solomon Islands, SW Pacific; crustal accretion and growth within an intra-oceanic setting. Tectonophysics 301(1–2):35–60. doi:10.1016/S0040-1951(98) 00214-5 Petterson MG, Cronin SJ, Taylor PW, Tolia D, Papabatu A, Toba T, Qopoto C (2003) The eruptive history and volcanic hazards of Savo, Solomon Islands. Bull Volcanol 65(2–3):165–181 Phinney EJ, Mann P, Coffin MF, Shipley TH (2004) Sequence stratigraphy, structural style, and age of deformation of the Malaita accretionary prism (Solomon arc–Ontong Java Plateau convergent zone). Tectonophysics 389(3–4):221–246. doi: 10.1016/j.tecto.2003.10.025
801 Rapp RP, Shimizu N, Norman MD, Applegate GS (1999) Reaction between slab-derived melts and peridotite in the mantle wedge: experimental constraints at 3.8 GPa. Chem Geol 160:335–356. doi:10.1016/S0009-2541(99)00106-0 Richards JP, Kerrich R (2007) Special paper: adakite-like rocks: their diverse origins and questionable role in metallogenesis. Econ Geol 102(4):537–576. doi:10.2113/gsecongeo.102.4.537 Rohrbach A, Schuth S, Mu¨nker C, Ballhaus C (2003) Island arc picrites from the Solomon Islands—origin by mantle matrix collapse. Geophys Res Abstr 5:11716 Royse KR, Kempton PD, Darbyshire DPF (1998) Procedure for the analysis of rubidium-strontium and samarium-neodymium isotopes at the NERC Isotope Geosciences Laboratory. NERC Isotope Geosciences Laboratory Report Series No. 121. NERC Isotope Geosciences Laboratory, Nottingham Schuth S, Rohrbach A, Mu¨nker C, Ballhaus C, Garbe-Scho¨nberg D, Qopoto C (2004) Geochemical constraints on the petrogenesis of arc picrites and basalts, New Georgia Group, Solomon Islands. Contrib Mineral Petrol 148(3):288–304. doi:10.1007/s00410004-0604-0 Sisson TW, Grove TL (1993) Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contrib Mineral Petrol 113(2):143–166. doi: 10.1007/BF00283225 Smith DJ (2008) From slab to sinter: the magmatic-hydrothermal system of Savo volcano, Solomon Islands. Ph.D. thesis, University of Leicester Stanton RL (1994) Ore elements in Arc Lavas. Oxford University Press, Oxford Taylor B (1987) A geophysical survey of the Woodlark-Solomons region. In: Taylor B, Exon NF (eds) Marine geology, geophysics and geochemistry of the Woodlark Basin–Solomon Islands, vol 7. Circum-Pacific Council, Houston, pp 25–48 Taylor B, Exon NF (1987) An investigation of ridge subduction in the Woodlark–Solomons region: introduction and overview. In: Taylor B, Exon NF (eds) Marine geology, geophysics and geochemistry of the Woodlark Basin–Solomon Islands, vol 7. Circum-Pacific Council, Houston, pp 1–24 Thirlwall MF (2002) Multicollector ICP-MS analysis of Pb isotopes using a 207Pb-204Pb double spike demonstrates up to 400 ppm/ amu systematic errors in Tl-normalization. Chem Geol 184(3– 4):255–279. doi:10.1016/S0009-2541(01)00365-5 Thorkelson DJ, Breitsprecher K (2005) Partial melting of slab window margins: genesis of adakitic and non-adakitic magmas. Lithos 79(1–2):25–41. doi:10.1016/j.lithos.2004.04.049 Tischendorf G, Forster HJ, Gottesmann B, Rieder M (2007) True and brittle micas: composition and solid-solution series. Mineral Mag 71:285–320. doi:10.1180/minmag.2007.071.3.285 Woodhead JD, Eggins SM, Johnson RW (1998) Magma genesis in the New Britain island arc: further insights into melting and mass transfer processes. J Petrol 39(8):1641–1668. doi:10.1093/ petrology/39.9.1641 Xiong XL, Xia B, Xu JF, Niu HC, Xiao WS (2006) Na depletion in modern adakites via melt/rock reaction within the sub-arc mantle. Chem Geol 229(4):273–292. doi:10.1016/j.chemgeo. 2005.11.008
123
